Identification and quantification of transposable element transcripts using Long-Read RNA-seq in Drosophila germline tissues

Transposable elements (TEs) are repeated DNA sequences potentially able to move throughout the genome. In addition to their inherent mutagenic effects, TEs can disrupt nearby genes by donating their intrinsic regulatory sequences, for instance, promoting the ectopic expression of a cellular gene. TE transcription is therefore not only necessary for TE transposition per se but can also be associated with TE-gene fusion transcripts, and in some cases, be the product of pervasive transcription. Hence, correctly determining the transcription state of a TE copy is essential to apprehend the impact of the TE in the host genome. Methods to identify and quantify TE transcription have mostly relied on short RNA-seq reads to estimate TE expression at the family level while using specific algorithms to discriminate copy-specific transcription. However, assigning short reads to their correct genomic location, and genomic feature is not trivial. Here we retrieved full-length cDNA  (TeloPrime, Lexogen) of Drosophila melanogaster gonads and sequenced them using Oxford Nanopore Technologies. We show that long-read RNA-seq can be used to identify and quantify transcribed TEs at the copy level. In particular, TE insertions overlapping annotated genes are better estimated using long reads than short reads. Nevertheless, long TE transcripts (> 4.5 kb)  are not well captured. Most expressed TE insertions correspond to copies that have lost their ability to transpose, and within a family, only a few copies are indeed expressed. Long-read sequencing also allowed the identification of spliced transcripts for around 107 TE copies. Overall, this first comparison of TEs between testes and ovaries uncovers differences in their transcriptional landscape, at the subclass and insertion level.